What about using the switching function along the high collision frequency?
I saw that the the default for amber is to use the upper boundary for the
switch function is qmcut and the lower boundary is the high boundary minus
2. Would it be possible to set the switch upper boundary to qmcut + 1 and
the lower boundary to qmcut -1?
Thank you
Fabrício
2013/8/2 Andreas Goetz <agoetz.sdsc.edu>
> Jason's interpretation is correct, the adaptive QM/MM code is currently
> available to developers collaborating with us. I had planned to release it
> with AMBER 12 but did not have time to properly test the code.
>
> I would strongly advice against using a step function at the QM/MM
> boundary. In the very least one should use a switching function for the
> forces as in Rode's "hot spot" method (Chem Phys 211, 1996, 313-323).
> Although it is possible to keep simulations somewhat under control with
> Langevin dynamics and a very high collision frequency (large value of
> gamma_ln) and still obtain structural properties (for example RDFs of ions
> in water) that are comparable to adQM/MM simulations with Bulo's DAS
> method. See for example my talk from the 2011 ACS spring meeting
> http://awgoetz.de/data/agoetz_adqmmm_acs-anaheim_2011.pdf (slides 18/19
> for step function, slides 28/29 for adQM/MM).
>
> All the best,
> Andy
>
> --
> Dr. Andreas W. Goetz
> Assistant Project Scientist
> San Diego Supercomputer Center
> Tel : +1-858-822-4771
> Email: agoetz.sdsc.edu
> Web : www.awgoetz.de
>
> On Jul 31, 2013, at 8:21 PM, Jason Swails wrote:
>
> > On Wed, Jul 31, 2013 at 3:05 PM, Fabrício Bracht <fabracht1.gmail.com
> >wrote:
> >
> >> HI Ross. I just read the paper describing the use adQM/MM in amber. It
> says
> >> there that the code has been implemented in amber, but you mentioned
> >> earlier that it will be a part of the next version of amber. I am
> guessing
> >> that the code that has been implemented is only available to developers,
> >> correct?
> >>
> >
> > The way I would interpret Ross's email is that adaptive QM/MM is only
> > developers. Variable solvent QM/MM is available in Amber 12.
> >
> >
> >> You also mentioned earlier the the discontinuity problems can be a
> problem
> >> if one does not use a very tight thermostat. By using langevin
> thermostat
> >> and a small gamma_ln value would I be able to use the adaptative scheme
> >> efficiently to include solvent effect on the qm region?
> >>
> >
> > The discontinuity is a problem of generic variable solvent -- a problem
> > that adaptive QM/MM solves by generating intermediate stages by which the
> > QM region is 'morphed' into MM continuously to avoid abrupt changes in
> the
> > potential that arises by swapping quantum water molecules. (At least
> that's
> > my understanding based on presentations I can remember).
> >
> > A small gamma_ln value would provide tighter temperature control and
> > _might_ stabilize vsolv=1 simulations. It may be worth trying.
> >
> > HTH,
> > Jason
> >
> > --
> > Jason M. Swails
> > BioMaPS,
> > Rutgers University
> > Postdoctoral Researcher
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Aug 01 2013 - 21:30:02 PDT
